#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/utils/OpAdapter.h"
namespace acl_op {
using npu_preparation = at_npu::native::OpPreparation;
using npu_utils = at_npu::native::NpuUtils;
namespace {
at::Tensor& pow_tensor_tensor_out_npu_nocheck(at::Tensor& result, const at::Tensor& self, const at::Tensor& exp)
{
at_npu::native::OpCommand cmd;
cmd.Name("Pow")
.Input(self)
.Input(exp)
.Output(result)
.Run();
return result;
}
at::Tensor& pow_tensor_scalar_out_npu_nocheck(at::Tensor& result, const at::Tensor& self, at::Scalar exp)
{
at_npu::native::OpCommand cmd;
if (exp.toFloat() == 2.0) {
cmd.Name("Square")
.Input(self)
.Output(result)
.Run();
} else {
cmd.Name("Pow")
.Input(self)
.Input(exp, self.scalar_type())
.Output(result)
.Run();
}
return result;
}
at::Tensor& pow_scalar_out_npu_nocheck(at::Tensor& result, at::Scalar self, const at::Tensor& exp)
{
at_npu::native::OpCommand cmd;
cmd.Name("Pow")
.Input(self, exp.scalar_type())
.Input(exp)
.Output(result)
.Run();
return result;
}
}
at::Tensor& pow_out(const at::Tensor& self, const at::Tensor& exp, at::Tensor& result)
{
auto output_size = op_infer::broadcast_ops_npu_output_size(self, exp);
npu_preparation::CheckOut(
{self, exp},
result,
self,
output_size);
if (!npu_utils::check_match(&result)) {
at::Tensor contiguous_result = npu_utils::format_contiguous(result);
pow_tensor_tensor_out_npu_nocheck(contiguous_result, self, exp);
npu_utils::format_fresh_view(result, contiguous_result);
} else {
pow_tensor_tensor_out_npu_nocheck(result, self, exp);
}
return result;
}
at::Tensor& pow_out(const at::Tensor& self, const at::Scalar& exp, at::Tensor& result)
{
npu_preparation::CheckOut(
{self},
result,
self);
auto exp_value = exp.toFloat();
if (exp_value == 0.0) {
return result.fill_(1);
} else if (exp_value == 1.0) {
return result.copy_(self);
}
if (!npu_utils::check_match(&result)) {
at::Tensor contiguous_result = npu_utils::format_contiguous(result);
pow_tensor_scalar_out_npu_nocheck(contiguous_result, self, exp);
npu_utils::format_fresh_view(result, contiguous_result);
} else {
pow_tensor_scalar_out_npu_nocheck(result, self, exp);
}
return result;
}
at::Tensor& pow_out(const at::Scalar& self, const at::Tensor& exp, at::Tensor& result)
{
npu_preparation::CheckOut(
{exp},
result,
exp);
if (!npu_utils::check_match(&result)) {
at::Tensor contiguous_result = npu_utils::format_contiguous(result);
pow_scalar_out_npu_nocheck(contiguous_result, self, exp);
npu_utils::format_fresh_view(result, contiguous_result);
} else {
pow_scalar_out_npu_nocheck(result, self, exp);
}
return result;
}
at::Tensor pow(const at::Tensor& self, const at::Tensor& exp)
{
auto output_size = op_infer::broadcast_ops_npu_output_size(self, exp);
at::Tensor result = npu_preparation::apply_tensor(self, output_size);
pow_tensor_tensor_out_npu_nocheck(result, self, exp);
return result;
}
at::Tensor pow(const at::Tensor& self, const at::Scalar& exp)
{
auto result_type = at::result_type(self, exp);
at::Tensor result = npu_preparation::apply_tensor(self, self.options().dtype(result_type));
auto exp_value = exp.toFloat();
if (exp_value == 0.0) {
return result.fill_(1);
} else if (exp_value == 1.0) {
return result.copy_(self);
}
at::Tensor self_copy = (self.scalar_type() != result_type) ? at_npu::native::custom_ops::_npu_dtype_cast(self, result_type) : self;
pow_tensor_scalar_out_npu_nocheck(result, self_copy, exp);
return result;
}
at::Tensor pow(const at::Scalar& self, const at::Tensor& exp)
{
auto result_type = at::result_type(exp, self);
at::Tensor result = npu_preparation::apply_tensor(exp, exp.options().dtype(result_type));
at::Tensor exp_copy = (exp.scalar_type() != result_type) ? at_npu::native::custom_ops::_npu_dtype_cast(exp, result_type) : exp;
pow_scalar_out_npu_nocheck(result, self, exp_copy);
return result;
}
at::Tensor& pow_(at::Tensor& self, const at::Tensor& exp)
{
acl_op::pow_out(self, exp, self);
return self;
}
at::Tensor& pow_(at::Tensor& self, const at::Scalar& exp)
{
acl_op::pow_out(self, exp, self);
return self;
}
}